Descripción del proceso

While we were able to follow Sharpless’s method for producing azidomethyl pivalate, we did encounter one quirk which is worth mentioning. The conditions call for the two reagents (sodium azide, chloromethyl pivalate) are added to room temperature water and then heated to 90 ˚C overnight, which would hint that the order of addition at room temperature would have little effect on the outcome. However, the order of addition is crucial to success. Suspending the chloromethyl pivalate in water, and subsequently adding the sodium azide, will result in ~90% yields, while forming a solution of azide in water first generally yields <10%.

We were able to reproduce Li’s ruthenium catalyzed Huisgen cycloaddition on his original diketone (36), in which the two cycloaddition partners are stirred in DMF at 90 ˚C overnight. However, when we submitted acetylene diamide (53), acetylene diethyl ester (52), and acetylene dicarboxylic acid (51) to the same conditions, no reaction ensued. Adjusting the heat of the reaction and utilizing a higher catalyst loading also did not produce detectable product.

50

Uncatalyzed versions of the azide-alkyne cycloaddition are known, so we began to

investigate a variety of literature-reported conditions that did not include the use of the expensive ruthenium catalyst. At the very least, we reasoned, we would develop hands-on knowledge of the reaction without wasting this expensive reagent, and we might be able to remove it entirely.

Increasing the concentration of the reagents did not originally yield results, but heating to a slow reflux in ether, overnight, gave the triazole dicarboxylic acid 57 in a promising 50% yield.71 _{However, because both the incompletely reacted starting material and the product}

contained two carboxylic acids, neither distillation nor column chromatography presented a good means of separating them. Thus, 57 proved to be difficult to isolate in high purity.

Scheme 12: Deprotecting the methylpivalate as another approach towards divergent target (54)

With this slightly impure sample, we proceeded to investigate whether the Weinreb amide 54 could be prepared on this, hopefully more stable, species. The methods attempted above for the preparation of Weinreb amide, including thionyl chloride, oxalyl chloride, and in- situ deprotonation with triethylamine, again tended to yield a dark brown sludge or oil with generally messy NMRs and a no clear major product. Meanwhile, the methylpivalate group seemed to cleanly deprotect from the triazole when subjected to Sharpless’s deprotection

51

conditions, but the resultant molecule (59) was extremely water soluble and never appreciably isolated.

Submitting the acetylene diethyl ester to the same cycloaddition conditions gave a 42% conversion into (58). However, the diethyl ester has the advantage of being a liquid, as opposed to acetylene dicarboxylic acid which melts at 180 ˚C. Inspired by Bertozzi’s development of a catalyst-free azide-yne click reaction,72 I attempted an uncatalyzed Huisgen cycloaddition under neat conditions, heated to 70 ˚C. Full conversion to (58) was achieved in two hours.

Unfortunately, we had run out of our supply of acetylene diamide (43) at this point, and a string of failed preparations dissuaded us from attempting a neat cycloaddition between it and azidomethyl pivalate to form (54). We reasoned, however, that the diester (58) no longer had its reactive alkyne and might more cleanly transform into the Weinreb amide (54).

With a reliable supply of (58), we began investigating the next two synthetic transformations that were available: the deprotection of the methylpivalate to alkylate the triazole, and the preparation of the Weinreb amide from the diethyl ester.

We began by applying Sharpless’s deprotection method, a 1:1 mixture of 1 M NaOH and methanol stirred for 30 minutes, to (58). While we observed 70% deprotection of the

methylpivalate by NMR, we also saw the formation of methyl ester groups that indicated the transesterification of methanol and the ethyl ester. Because this product mixture was more difficult to purify, isolate, and characterize, we sought to remedy this issue by switching the solvent to ethanol, so that any transesterifications would reform the ethyl ester.

An identical deprotection in ethanol reproducibly produced complex mixtures with 1H- NMR indicating a mixture of products, or possible polymerization, that broadened the ethyl ester signals. Running the reaction without an alcohol cosolvent (i.e., in 1M NaOH) resulted in two

52

phases that did not react. Using distilled, dry ethanol and either sodium metal or lithium methoxide to generate sodium ethoxide or lithium ethoxide lead to decomposition of (58).

Still, the methylpivalate species should be more reactive than the ester, so we ran a battery of tests to try to thread the needle on reactivity. We tested various equivalents of

hydroxide, at varying concentrations, temperatures, and reaction times. Eventually, we were able to isolate (60), though the synthesis is not always reproducible. Our optimized method utilizes an excess (~2.5 equivalents per ester) of 1 M hydroxide, in a 1:1 mixture with ethanol, stirred for two hours.

We then submitted the naked-triazole ethyl-ester (60) to the conditions for alkylating the triazole ring: potassium carbonate and bromo-2-butyloctane stirred in DMF. While the

appropriate methylene signal indicated an N2-alkylated species on 1H-NMR (62), the ester peaks were consumed and the alkylated product was never isolated. It is probable that, as we were concerned it might, the nucleophilic triazole anion was a strong enough nucleophile to attack the esters in solution, resulting in an undesirable mixture of products.

Even though methylpivalate is supposed to deprotect under very mild conditions, the difficulties described above indicate that it is not wholly orthogonal from the ester. Thus, we decided to briefly investigate one-pot transformations that would leverage this shared reactivity. Removing all three ethers (including the methylpivalate) with a single nucleophile, like N,O- dimethylhydroxylamine (50), should produce the Weinreb amide as well as the deprotected triazole, as long as the deprotection of methylpivalate performs identically with non-hydroxide nucleophiles.

Combining (58) with 10 equivalents of (50) in dichloromethane (its extraction solvent) produced no reaction after three days of stirring, but combining the salt, (49), mixed with

53

triethylamine in methanol was more promising. The resulting solid was not soluble in chloroform, as is expected for deprotected triazoles, and an 1_{H-NMR in MeOD revealed the} formation of Weinreb amide and the removal of the t-butyl group characteristic of

methylpivalate.

Closer examination of the major product, however, revealed an additional methylene unit, likely connected to the triazole. Dhanak et al. report that methylpivalate sometimes fails to properly deprotect, and instead forms the methanol moiety expected from the simple cleavage of the t-butyl ester. 73 They recommend a heated solution of dilute KOH for stripping off the remaining functional group to expose the bare amine, but attempts to apply this ended in either no reaction or degradation of this delicate system.